Simple Question regarding analysis in Closed Population

[Klimstra]

I did a small mammal mark-recapture study and used MARK to analyze my results. I am teaching myself the program because I do not have the resources here to learn from someone, so any help will be tremendous. I am concerned that I am not doing something right based on the results from the analysis. I chose to run my data assuming closed population and ran three different models (C=P, N same . . . C=P, N different. . . .and C diff., P diff., N. diff). I have extremely small sample sizes for a few of the species that I trapped and MARK gives me abundance estimates equal to the actual number captured which does not seem right. What am I doing wrong? I thought this could be a possibility since the only individuals captured in these scenarios were captured multiple times. Additionally, standard error for all of my analyses were rather strange when the sample sizes were extremely low. Thanks for an HELP.

RK

[dhewitt]

RK (I'm guessing Ryan),

You're in luck, for two reasons. One, there's a rather exhaustive and helpful "manual":

http://www.phidot.org/software/mark/docs/book/

You'll probably be most interested in the early chapters and Chapter 14 on closed captures models.

Second, you're at an institution with a long history in these methods, so help is right down the hall.

Unfortunately, based on the limited information you provided, my guess is that closed captures models will be poor and misleading for your data set. Whacky SEs are a pretty good indicator. It very well may be that you're doing things "right" in MARK, but MARK (or model fitting in any other program without strong priors in a Bayesian analysis) does not have enough information in your data set to produce sensible results. People always want to know abundance, but sometimes the data just are not sufficient to go there. I hope I am wrong about your situation.

- Dave

P.S. - Probably worth reviewing the FAQ as well:
viewtopic.php?f=39&t=9

[murray.efford]

First, I suggest you think hard about whether non-spatial models are up to the task. Some of us think they are no longer the method of choice for small-mammal trapping studies. The primary reason is that the target population is usually ill-defined (unless you trapped throughout a habitat island). Trapping on a grid also induces heterogeneity in individual capture probability; this is a headache in conventional models, but is handled automatically by spatially explicit models. Sample-size requirements are similar for spatial and non-spatial models. Introductory documentation for spatially explicit capture-recapture software (Density, secr, SPACECAP etc.) is not up to the high standard set by the Gentle Introduction, but users seem to manage (the Gentle Introduction is a good source for general concepts such as maximum likelihood and model selection).
Murray